Although it was always possible to set up XTC mode so that it technically works without problems, XTC mode always had two major problems:
a) CW never provided any serious information on how to set up the whole thing.
b) XTC has a horrible latency in the range of half a second that makes it impossible to use the stuff if this latency is not compensated by the host.
Since the release of Cubase SX 2.0 (and as far as I know, the latest Logic version) a) is not an issue anymore. All plugins, including the XTC ones, are compensated realtime all through the whole signal path.
However, b) is still a big problem. People in this forum keep asking questions like "how to set up XTCProject.pro" and the like, and it is definitely up to CW to clarify this.
But all in all, XTC is becoming an option now. One can even have sample accurate XTC delays as sends in SX 2.0.
Sequencer manufacturers help XTC mode
Your latency in XTC mode should be 2x your soundcard buffer settings + a tiny overhead from the internal dsp calculations. If you're getting 1/2 second latency without compensation then something is wrong.
Also, one of the problems with 'automatic compensation' across the whole audio engine is that its making all the additional buffering necessary to keep things 'compensated' transparent to the user. When individual tracks are 'compensated' their output can simply be shifted forward in time to allow the audio to exit the plugin chain/synth in correct time/phase with the rest of the project. However when bus/groups are used with dsp plugins and/or software plugins with a high overhead every other group & output is going to be 'buffered' to match. If you're using a older p4 (less than 2ghz) or a p3/athlon and start using inserts here & there without thought given to the additional buffers that are being created 'behind the scenes' then yes the audio engine will likely place a lot of addional strain on the machine's memory bus (and therefore cpu).
At least this is my understanding, I don't often use XTC mode often (nor do I own a UAD-1 anymore) so please feel free to correct anything I may have misstated.
Also, one of the problems with 'automatic compensation' across the whole audio engine is that its making all the additional buffering necessary to keep things 'compensated' transparent to the user. When individual tracks are 'compensated' their output can simply be shifted forward in time to allow the audio to exit the plugin chain/synth in correct time/phase with the rest of the project. However when bus/groups are used with dsp plugins and/or software plugins with a high overhead every other group & output is going to be 'buffered' to match. If you're using a older p4 (less than 2ghz) or a p3/athlon and start using inserts here & there without thought given to the additional buffers that are being created 'behind the scenes' then yes the audio engine will likely place a lot of addional strain on the machine's memory bus (and therefore cpu).
At least this is my understanding, I don't often use XTC mode often (nor do I own a UAD-1 anymore) so please feel free to correct anything I may have misstated.